Estimated incident cost savings in shipping due to inspections
The effectiveness of safety inspections has been analysed from various angles, but until now,relatively little attention has been given to translate risk reduction into incident cost savings.This paper quantifies estimated cost savings based on port state control inspections andindustry vetting inspections. It is based on a unique dataset of 515,194 ship arrivals andinspections from the United States of America and Australia, and inspections of threeindustry vetting inspection regimes, for the time period 2002 to 2007. The risk reducingeffect of inspections is estimated by means of duration models, in terms of inspection gains based on the probability of survival. The results suggest average total estimated cost savingsin the range of USD 74 to 192 thousand (median USD 19 to 46 thousand) owing to reducedrisk of total loss due to a port state control inspection. Cost savings for industry inspectionsare found to be even higher, especially for tankers. The savings vary by type, age and size ofthe ship. The benefits of an inspection are in general larger for older and larger vessels, andalso for vessels with undefined flags and unknown classification societies. As inspectioncosts are relatively low in comparison to potential cost savings, the results underline theimportance in determining high risk ships to prevent costs due to total loss of ships.
In principle, the estimated risk probabilities may also be of interest for theinsurance market (P&I Clubs
and marine underwriters).The effectiveness of port state control or industry inspections has been treated in theliterature from various angles, by Knapp and Franses , , , Carriou et al. , andPayoyo . Most emphasis has so far been put on risk reduction in the sense of decreasingthe probability of detention or casualty due to an inspection. Our aim in this paper is toquantify the decrease in costs associated with casualties.The paper has the following structure. Section 2 describes the dataset and explains thevarious components used for the calculation of the base values to determine incident costssavings. Section 3 contains econometric models to estimate the effect of inspections. Section4 presents the estimated incident cost savings due to port state control inspections, andSection 5 concludes.
2. Dataset and determination of base values for incident costs
The empirical analysis is based on data on an individual ship level of arrivals and port statecontrol inspections and is complemented by information that is required to determineestimated cost savings of incidents. Notwithstanding the political aspects of port state controldata on an individual ship level, we obtained the generous cooperation of the United StatesCoast Guard (USCG) and the Australian Maritime Safety Authority (AMSA). The datasetcontains 515,194 ship arrivals of 15,819 ships for the time period from 2002 to 2007. We usealso vetting inspection information from RightShip for dry bulk carriers and vettinginspections from OCIMF and CDI for oil and chemical tankers.The arrival and inspection dataset is merged with casualty data of the same time period(including relevant casualties from 2008), mainly originating from Lloyd’s Register Fairplay.The dataset is further complemented by earnings, which represent ship economic cycles,obtained from the Shipping Intelligence Network of Clarksons. The combination of thesevarious data sources allows us to estimate the probability of a ship having a casualty.Our aim is to translate the effect of a port state control inspection into a monetary value. Thisvalue represents the total estimated cost savings (abbreviated henceforth by TECS) owing tothe reduction of incident risk caused by a port state control inspection. These savings arecalculated on an individual ship level and depend on the average costs of casualties. Thesecosts are also denoted as “base values”. The literature on costs of incidents provides an ideaof the complexity of determining a realistic base value for incident costs. Wood identifies four components to the costs of marine incidents, that is, lost assets, loss of cargo,lost lives, and pollution. It is very complex to estimate each of these components. Talley investigates vessel damage cost differentials for some ship types and determinants of property damage costs of tanker accidents , and Goulielmos and Giziakis  consideruncompensated costs of marine accidents. For loss of life, various figures found in theliterature are summarized by Skjong . The values vary considerably, depending on theregion. The International Maritime Organization (IMO) uses a value of USD 1.5 million inits guidelines for Formal Safety Assessment (FSA) . In addition to the cost of lost lives,the costs of injuries are also important. The European Union Project SAFEDOR suggests a range of USD 20,000 to 70,000 per injury, while the IMO FSA methodology
Protection and Indemnity Clubs provide third party liability insurance in the shipping industry and are comprised of shipowners. Marine underwriters primarily insure hull and machinery.
4suggests a value of USD 42,000 per injury. As concerns the cost of pollution, there iscurrently no consensus at IMO level, but SAFEDOR provides an average figure of USD60,000  per ton. This figure does not take into account the associated costs ofenvironmental damage (such as loss of animal life) and of other socio-economic factors, seeGrigalunas
 and Grey , since it is very difficult to estimate these costs.In the rest of this section, we consider three alternative base values to calculate TECS, that is,historical claims figures, insurance premiums, and the total insured value.In principle, historical claim figures are available from P&I Clubs and Marine Underwriters.The P&I Club figures cover all third party liability (general average, pollution, personnel,third party property damages) but exclude claims of cargo interest. The underwriter figuresrepresent claims with respect to hull and machinery. Both values, however, will only reflectclaims above the ship owner’s deductible
, which can vary depending on the ship owner performance. In addition, the claim figures already reflect the effect of port state control andother inspections since they are historical figures influenced by safety inspections that reducerisk. The claim figures are therefore discounted. Average values are obtained from Knapp and from P&I Clubs and Marine Underwriters .The insurance premium reflects the ship owners’ portion of the value at risk as perceived bythe insurance market. It excludes so-called cargo interests. Cargo interest premiums are verydifficult to obtain since they are determined in a highly competitive global market andinformation on premiums is commercially sensitive. Third party liability will, however,cover ship owners’ liabilities for third party claims for cargo. Similar to the claims, the premium only reflects coverage above the deductible, which can vary considerably becauseof individual judgment and agreed underwriting guidelines of cargo underwriters. We haveidentified premiums per ship type from Drewry Shipping Consultants  based on theInternational Group of P&I Clubs
, adding 20% for administrative costs .The most comprehensive base value to estimate cost savings of incidents is the total insuredvalue (abbreviated henceforth by TIV). This value is influenced by the liability environmentembedded in the legislative framework of the shipping industry. The components of TIV arethe same as the ones identified by Wood  for the total costs of maritime incidents. Thisdetermines the value that can be recovered by insurance or stated otherwise, the value thatcan be insured. Based on insurance cover, the components of TIV are the following: (1) costof hull and machinery (insured by Marine underwriters), (2) third party liability coverage(insured by P&I Clubs), (3) oil pollution coverage for oil tankers (above the insurance limitsfrom P&I Clubs), and (4) cargo values for cargo carrying vessels (for passenger vessels, thisis replaced by liability limits for injury or death of passengers).First we consider the costs of hull and machinery. These values are insured on a valued policy basis according to the United Kingdom Marine Insurance Act of 1906. This meansthat the insured value is agreed between the underwriter and the owner and the underwriterwill use his expert knowledge and other tools such as the Shipping Intelligence Network ofClarkson’s  to assess the value. In addition, most vessels are mortgaged, which meansthat the insurance policy is assigned to the bank financing the vessel. The bank’s primaryconcern is to ensure that the insured value is sufficient to cover the mortgage on the vessel inevent of a total loss. The loan agreement will therefore stipulate the minimum value that is
The deductible is the ship owner’s portion of the claim; according to Knapp , based on industry sources, the deductiblecan vary from USD 50,000 to 250,000 for hull and machinery, from USD 5,000 to 30,000 for personnel related claims, andfrom USD 10,000 to 100,000 for all other claims.
The International Group of P&I clubs covers about 90% of the world fleet by gross tonnage, see Drewry .
5acceptable to the bank. These values can fluctuate according to the market situation. For the purpose of this article, we use the second hand prices of vessels ($/DWT
) from theShipping Intelligence Network of Clarkson’s , which provides monthly time series andwhich are adjusted for inflation
. The second hand prices are used since these prices providethe most realistic value of the asset. Next, we consider third party liability coverage. This value is specified as being USD 10million per incident , which in our case represents the maximum lower pooled amountfor an individual P&I Club (the split up is USD 8 million plus an allocated amount for theowner’s deductible and a portion of excess loss reinsurance). According to industry sources,the trend is upwards from USD 8 million to USD 10 million. According to the InternationalGroup of P&I Clubs, for oil tankers and for the period 1978-2002, 98% of the cases arewithin the current limit of USD 8 million . We will therefore take the USD 10 millionlimit including an allocation for the self-insured amount as our basis for the analysis.Third, we consider oil pollution limits. These limits are based on the 1992 Civil LiabilityConvention and the 1992 Fund Convention and Supplementary Fund Protocol . Theselimits for oil tankers are in excess to the third party liability used for the P&I Clubs. Thelimits are presented in Table 1. The oil pollution limits are expressed in terms of so-calledspecial drawing rights (SDR’s) and conversion rates are defined by the InternationalMonetary Fund (IMF). For the conversion of SDR’s into USD, we used daily conversionrates
from January 2000 to December 2007. For other ship types than oil tankers, IMOadopted the International Convention on Civil Liability for Bunker Oil Pollution in March2001, which has come into force in November 2008. This aspect is not taken into account inour analysis, as our data period runs until the end of 2007.
Table 1: Oil pollution limits for oil tankers
Ship size (in gt) Till Oct 31, 2003 From Nov 1, 2003
up to 5000 3,000,000 SDR(USD 4.6 million)4,510,000 SDR(USD 6.9 million)5000 to 140,0003,000,000 SDR plus 420 SDRfor each additional tonnage(USD 4.6 million plus USD 647)4,510,000 SDR plus 631 SDR for eachadditional tonnage(USD 6.9 million plus USD 972)140,000 and above59,700,000 SDR(USD 92 million)89,770,000 SDR(USD 138.3 million)
Note: The amounts in USD are approximate values as of December 2008.
Finally, the fourth component of TIV is the value of cargo that is carried by ships each year.We use data from the United Nations Commodity Trade Statistics database , togetherwith an estimated share of seaborne cargo varying from 64% in 2000 to 70% in 2007. Therelevant value is calculated for each year. The total seaborne values are compared withvalues from the UNCTAD Maritime Transport Review , which represent cargo value ofimports (c.i.f.), and to values given by Hoffman  based on data from GlobalInsight. Weuse the average of the three values as the base to determine average cargo values per cargocarrying capacity (DWT) per day. The resulting cargo values are presented in Table 2. As anestimate of total days at risk, we use an average of 224 days at sea, based on data used in theIMO 2009 Greenhouse Gas study . It is worth noting that these values do not reflect th